Nicotinic Regulation of Energy Homeostasis

Abstract

Introduction:

The ability of nicotine, the primary psychoactive substance in tobacco smoke, to regulate appetite and body weight is one of the factors cited by smokers that prevents them from quitting and is the primary reason for smoking initiation in teenage girls. The regulation of feeding and metabolism by nicotine is complex, and recent studies have begun to identify nicotinic acetylcholine receptor (nAChR) subtypes and circuits or cell types involved in this regulation.

Discussion:

We will briefly describe the primary anatomical and functional features of the input, output, and central integration structures of the neuroendocrine systems that regulate energy homeostasis. Then, we will describe the nAChR subtypes expressed in these structures in mammals to identify the possible molecular targets for nicotine. Finally, we will review the effects of nicotine and its withdrawal on feeding and energy metabolism and attribute them to potential central and peripheral cellular targets.

Introduction

The ability of nicotine, the primary psychoactive substance in tobacco smoke, to regulate appetite and body weight is one of the factors cited by smokers that prevents them from quitting and is the primary reason for smoking initiation in adolescents, and in particular, teenage girls. The regulation of feeding and metabolism by nicotine is complex, and recent studies have begun to identify nicotinic acetylcholine receptor (nAChR) subtypes and circuits or cell types involved in this regulation. Recently, a conceptualization of the regulation of feeding and energy metabolism that is gaining wide acceptance has proposed the existence of two complex and partially interacting brain circuits: a homeostatic system centered on the hypothalamus and a hedonic system centered on the cortico-limbic-striatal circuits. Given the ability of nicotine acting through nAChRs to regulate both these systems, it is relevant to evaluate the role of these interconnected systems in the ability of nicotine to regulate food intake and metabolism.

In this review, we will briefly outline the effects of nicotine on body weight and energy metabolism and then discuss the possible cellular and molecular targets that may mediate these effects. Nicotine targets in systems regulating energy homeostasis will be subdivided into peripheral input and output structures, and central integrative structures (Figures 1 and 2). With respect to central structures, we will discuss how nicotine affects homeostatic and reward circuits that regulate feeding and energy metabolism, as well as their interactions. Drug dependence is known to involve appetitive mechanisms, and the effects of nicotine on body weight, eating, and obesity are therefore likely to contribute to nicotine addiction. Starting from pioneering studies in the 1970s and early 1980s (Falkeborn, Larsson, & Nordberg, 1981; Grunberg, 1982; Grunberg, Bowen, Maycock, & Nespor, 1985; McNair & Bryson, 1983; Schechter & Cook, 1976; Wack & Rodin, 1982), multiple effects of nicotine and smoking on food consumption, energy expenditure, as well as food hedonics have been identified. In this review, we will outline the cellular and molecular mechanisms that may underlie the effects of nicotine on energy intake, expenditure, and hedonics of food consumption.

Figure 1.

Schematic representation of the principal structures of the systems that regulate energy metabolism. AMY = amygdala; ARC = arcuate nucleus; DMH = dorsomedial hypothalamic nucleus; GI = gastrointestinal; INS = insula; LH = lateral hypothalamus; nAc = nucleus accumbens; NTS = nucleus of the solitary tract; OFC = orbitofrontal cortex; PFC = prefrontal cortex; PVH = paraventricular hypothalamic nucleus; VMH = ventromedial hypothalamic nucleus; VTA = ventral tegmental area.

Figure 2.

Expression of nicotinic acetylcholine receptors in systems that regulate energy metabolism. For abbreviations, see legend to figure 1. ANS = autonomic nervous system; BAT = brown adipose tissue; ENS = enteric nervous system; WAT = white adipose tissue.

Effects of Nicotine on Body Weight and Metabolism

Due to the almost ubiquitous expression of nAChRs in central and peripheral neuronal cells and nonneuronal cells in peripheral organs that are involved in the regulation of body weight, nicotine has potentially complex actions on energy homeostasis. Thus, the impact of systemic nicotine may be different depending on food quality and availability, behavioral or metabolic states as well as individual genetics, personality, and habits.

Moreover, it is well known that nicotine has an inverted U dose-response relationship on its receptors and in its behavioral actions, and it desensitizes as well as upregulates its receptors upon prolonged administration. Therefore, nicotine dose and modality of administration (e.g., acute, chronic repeated, or chronic continuous administration, passive, or active administration) may lead to markedly different effects.

Finally, it must be kept in mind that averaged data from laboratory, often inbred, animals kept in impoverished environmental conditions and eating lab chow food underestimate the varieties of impacts that nicotine may exert in animals living in natural environments and humans.

In general terms, energy homeostasis can be regulated by intervening at the level of food (i.e., energy) intake and/or on energy expenditure. Food intake depends on meal size and frequency as well as absorption of nutrients, whereas energy expenditure is thought to consist of basal metabolic rate, adaptive thermogenesis, and energy cost of physical activity (activity thermogenesis).

In laboratory rodents freely fed on a standard diet (around 5% fat), acute and chronic nicotine treatments result in a decrease in body weight gain that is associated with one or more of these effects: (a) decreased food intake due to reduction of meal size (i.e., increased satiation) (see e.g., Bellinger, Wellman, Harris, Kelso, & Kramer, 2010; Blaha, Yang, Meguid, Chai, & Zadák, 1998; Bray, 2000; Grunberg, Bowen, & Winders, 1986); (b) increased energy metabolism (Schechter & Cook, 1976; Sztalryd, Hamilton, Horwitz, Johnson, & Kraemer, 1996); (c) increased lipolysis (Andersson & Arner, 2001); or (d) increased physical activity (Gillman, Kosobud, & Timberlake, 2008, 2010). Indeed, using indirect methods, it was shown that nicotine can lower body weight set point in both rodents (Frankham & Cabanac, 2003) and humans (Cabanac & Frankham, 2002).

Similar results have been obtained with short/long-term exposure to cigarette smoke in rodents (Chen et al., 2007; Wager-Sdar, Levine, Morley, Hoidal, & Niewoehner, 1984), suggesting that nicotine is the main agent responsible for the effects of tobacco smoke on energy homeostasis. Accordingly, in humans, tobacco smoking can reduce food intake and increase physical activity and metabolic rate, whereas cessation of smoking leads to hyperphagia and weight gain (Filozof, Fernández Pinilla, & Fernández-Cruz, 2004).

The widespread expression of nAChRs makes it likely that several, not mutually exclusive, nicotine targets, both central and peripheral, some in series and some in parallel, mediate nicotine effects on energy homeostasis. Although nicotine effects on specific portions of the regulatory systems of energy homeostasis are starting to be understood, the wide variety of nicotine-elicited processes that result in a positive or negative energy homeostatic state continues to make it difficult to determine what receptor(s) and mechanisms(s) prevail after nicotine administration in different pathophysiological states.

In the following sections, we will review what nAChRs are expressed and how they can affect the different components of the energy homeostasis system, distinguishing nicotine effects on peripheral and central structures that control energy homeostasis.

In this context, some preliminary methodological considerations should be made: The availability of knockout animals for nAChR subunit has allowed a more precise assessment of the specificity of the tools used in nicotine research. As a result of this approach, it has become clear that the vast majority of the antibodies against nAChR subunits used in immunohistochemical studies are nonspecific (Jones & Wonnacott, 2005; Moser et al., 2007). nAChR subtypes with complex composition are expressed in many tissues, but subtype selective pharmacological tools are relatively limited, in particular ligands with selectivity for uncommon nAChR subtypes. Further, nonquantitative polymerase chain reaction (PCR) has often been used to assess tissue expression of nAChR subunits, a method that does not faithfully reflect the amount of target mRNA. In addition, it should be considered that mRNA is not always translated into protein, expression of individual nAChR subunits does not always reflect assembled and functional receptor, the pool of assembled receptors is often expressed in intracellular compartments rather than on the plasma membrane, and receptors expressed on the cell membrane may be in functional or nonfunctional states. Molecular genetics approaches leading to inactivation or hyperexpression of selected genes in specific cell types have advantages in the study of complex systems and effects, since they allow in vivo investigations of organs and the entire animal. Yet, the use of these methods in the field of nicotine and energy homeostasis has been very limited till now.

Peripheral Structures

Information relevant to energy metabolism (meal features, food digestion, nutrient levels, and adiposity) is sent to the brain through humoral signals or peripheral sensory nerves. On the other hand, information output from the brain to energy metabolism effectors is sent mainly through peripheral motor autonomic or somatic nerves. In addition, the enteric nervous system, which is independent, though modulated by motor autonomic nerves, has the capability to regulate digestive processes independently. All these neuroendocrine structures express nAChRs and may be directly influenced by nicotine. Finally, effector organs such as white (WAT) or brown (BAT) adipose tissues can also express nAChRs and be targets for nicotine.

Neuronal Inputs

The main type of neurotransmission in primary sensory neurons is glutamatergic, yet nAChRs are expressed in most cells constituting the peripheral branches of sensory systems. Accordingly, nicotine can modulate multiple sensory systems at multiple levels, including several afferents that affect ingestive processes.

Taste Afferents

At concentrations similar to those attained in smokers, nicotine has prominent effects on taste responses (Megerdichian, Rees, Wayne, & Connolly, 2007). Nicotine has a bitter taste and may influence food intake through a direct action on taste receptors and/or taste pathways. Transient receptor potential ion channel M5 is expressed in taste receptor cells and is necessary for the transduction of prototypical bitter tastants, such as quinine. Its genetic deletion decreases nicotine responses in primary sensory neurons, showing that nicotine shares a common transduction mechanism with common bitter tastants. In addition, and in an independent manner, nicotine taste is mediated by nAChRs expressed in the primary neuron and/or the taste receptor cell (Oliveira-Maia et al., 2009). Accordingly, immunoreactivity of α3, α4, α5, α7, and β2 subunits has been detected in lung bitter taste receptor cells (Dehkordi et al., 2010), but some caution should be used for the specificity of the antibodies used, see above.

Nociceptors

Nicotine is an irritant that stimulates nociceptors (Megerdichian et al., 2007; Rau, Johnson, & Cooper, 2005; Simons, Boucher, Carstens, & Carstens, 2006). Stimulation by chemical irritants like nicotine at the level of the tongue excites trigeminal afferents and spinal trigeminal nucleus, whereas stimulation at the level of the throat excites IX and X nerve afferents and the nucleus of the solitary tract (NTS) (Boucher et al., 2003). Most nociceptor classes express functional nAChRs (Rau et al., 2005). α7 nAChRs may be restricted to C afferents, whereas heteromeric nAChRs are expressed by both C and Aδ afferents (Rau et al., 2005). Dorsal root ganglia (DRG) express a majority of α3β4* and α6β4* and a minority of α3/α6β2* nAChRs (Hone, Meyer, McIntyre, & McIntosh, 2011), and functional studies indicate that around 50% of nociceptive neurons in these ganglia express α3/α6β2-sensitive nAChRs (Spies, Lips, Kurzen, Kummer, & Haberberger, 2006).

In addition, nicotine affects pain transmission elicited by other stimuli. Non-β2* nAChRs expressed in nociceptors have an antinociceptive function (Yalcin et al., 2011). Moreover, nAChRs exert an antinociceptive action at spinal and supraspinal levels. Genetic and pharmacological approaches have shown that α4β2 nAChR mediates nicotine-mediated supraspinal antinociception (Damaj et al., 2007). At the spinal level, nicotine exerts a complex modulatory action on pain and analgesia. A tonic cholinergic tone through β2* nAChRs and GABA transmission increases nociceptive threshold for both mechanical and thermal stimuli (Yalcin et al., 2011). Other spinal pain control mechanisms are not mediated by β2* nAChRs (Cordero-Erausquin & Changeux, 2001). Accordingly, nicotine pharmacological analgesic action at spinal level is mediated by both α4β2 and other subtypes, possibly including α7 (Damaj et al., 2007).

Visceral Afferents

nAChRs are expressed in visceral vagal afferents to the caudal NTS, as well as in the efferent parasympathetic motoneurons of the dorsal motor nucleus of the vagus nerve (DMnX), and in the sympathetic afferents to the DRG. In the nodose ganglion, binding to nicotinic ligands amounts to about 10% of the binding in autonomic motor ganglia, being 85% of them β4* receptors and 15% β2* receptors. The subtype composition of these receptors is very similar to those expressed in autonomic motor ganglia, since α3β4 accounts for 50%, α3α5β4 for 35%, and α3β2β4 for 15% of receptors (Mao, Yasuda, Fan, Wolfe, & Kellar, 2006).

Nucleus of the Solitary Tract

The NTS, a complex brain stem nucleus located in the dorsal medulla oblongata, plays a prominent role as an input station to energy metabolism control systems (Grill & Hayes, 2009). Its rostral part receives taste afferents, whereas its intermediate portion receives gastrointestinal afferents. In addition, it receives projections from second-order neurons of the dorsal horn (spinal sympathetic afferents) and spinal trigeminal nucleus (trigeminal afferents), as well as humoral signals directly or indirectly through the area postrema (AP). AP, an area devoid of blood brain barrier (BBB), is located just dorsal to the NTS and sends its projections mainly to the NTS. From a functional standpoint, the AP is considered to be a chemosensor associated with the NTS. The NTS/AP complex sends heavy projections either directly or indirectly through the parabrachial nucleus to upper centers including hypothalamic nuclei of the homeostatic system.

As described earlier, sensory afferents to NTS express a diverse array of nAChR subtypes. In addition, in both rostral and caudal NTS, intrinsic neurons receive cholinergic innervation (both interneurons and afferents from adjacent regions, such as DMnX) and express nAChRs, both α7 and non-α7 receptors of unknown composition (Smith & Uteshev, 2008; Uteshev & Smith, 2006). Accordingly, at this level, nicotine powerfully modulates sensory afferents and their integration. For instance, nicotine acting at the level of the tongue is able to suppress taste responses in the NTS. This action is thought to be mediated by nAChRs expressed by trigeminal nociceptors that, in turn, inhibit taste afferents in the NTS (Simons et al., 2006).

In addition to local integration and relay to upper centers, visceral afferents reach the caudal NTS and form reflex arcs. A primary category of reflexes is vago-vagal reflexes that regulate several aspects of gastrointestinal, mainly gastric, function. Nicotine elicits a vago-vagal reflex that leads to decreases in intragastric pressure and fundus tone in the rat, thus mimicking the passage of food in the esophagus (Ferreira et al., 2002, 2005). This effect has been shown to be mediated by α4β2 nAChRs, probably expressed on vagal afferents, and to involve a reflex circuit that comprises NTS noradrenaline neurons and DMnX efferents. This reflex potentiates satiation and delays food transit.

Although nAChRs in the NTS and related circuits have been widely investigated, the complexity of this nucleus still prevents a precise reconstruction of the nAChR subtype expression pattern and functional role.

In conclusion, nicotine has two categories of effects on peripheral sensory processes. First, nicotine is a sensory cue by itself. As a sensory cue, it can develop incentive value through association with reinforcers such as palatable foods and nicotine itself. Therefore, nicotine-related sensory cues can enter the complex chain of positive and negative incentive stimuli that regulate food preference and feeding behaviors (see below). Second, nicotine alters the processing of other sensory cues including those related to foods and food processing. This may alter the information on food quantity and quality for homeostatic responses and on food palatability for reward processing.

Humoral Inputs

Humoral signals include meal control signals from the gastrointestinal tract, blood nutrient levels that are used by brain circuits as measures of the metabolic status of the body (Sanchez-Lasheras, Könner, & Brüning, 2010), and adiposity signals from the WAT (the body energy stores).

Meal control signals include gastric and intestinal wall distension and hormones released from enteroendocrine cells (Castaneda, Tong, Datta, Culler, & Tschöp, 2010; Chaudhri, Salem, Murphy, & Bloom, 2008). The former is recorded by stretch receptors in the digestive system and is transmitted through the vagus and sympathetic nerves to the brain stem (Cummings & Overduin, 2007). Enteroendocrine cells associated with the gastrointestinal tract are modulated by neuronal or hormonal signals, sense nutrient levels and taste of the gastrointestinal content through a wide array of G-protein coupled receptors (Geraedts, Troost, & Saris, 2011), and produce short-lived hormones that promote appetite (ghrelin, from the gastric fundus and, to a lower extent, small intestine) or satiation (the principal hormones are cholecystokinin (CCK), glucagon-like peptide 1 (GLP-1), peptide YY (PYY), and serotonin (5HT) from small intestine). Note that in this context, satiation refers to the processes involved in meal termination whereas satiety refers to the postprandial absence of hunger, which contributes to the determination of the interval between meals. Overall, release of these hormones depends on meal content and is broadly related to energy content but also depends on meal quality (Cummings & Overduin, 2007; Young, 2011). These hormones can act directly on receptors expressed by hypothalamic and/or brain stem neurons and/or indirectly on receptors expressed on vagus nerve terminals that project to the NTS.

Acute or chronic nicotine treatment can change RNA expression and plasma levels of several enteroendocrine peptides (Chowdhury, Hosotani, Chang, & Rayford, 1990; Chowdhury, Hosotani, & Rayford, 1989; Gomez et al., 1996; Wong & Ogle, 1995). The existence of direct effects on enteroendocrine cells has not been extensively investigated. nAChRs stimulate 5HT release from enterochromaffin cells (Racke, Reimann, Schwörer, & Kilbinger, 1996), whereas no evidence for release of histamine or gastrin was obtained (Brenna et al., 1993; Matsuno, Matsui, Iwasaki, & Arakawa, 1997). Localization of nAChR immunoreactivity in the bowel mucosa showed positivity in a subset of enterochromaffin cells, and no signal in any other kind of enteroendocrine or nonendocrine cell (Kirchgessner & Liu, 1998). Overall, it is likely that nicotine-induced alterations of enteroendocrine hormones are largely indirect, mediated by the autonomic and enteric nervous systems and changes in food intake and enteric motility.

Adiposity signals are tonically released in proportion to total body fat, the primary body energy stores. The principal adiposity signals are insulin, from pancreatic β cells, and leptin, mostly from WAT cells. Insulin serum levels are proportional to adiposity, since responsiveness of the β cells to glucose is proportional to body fat (Woods & Seeley, 2000). Leptin is secreted in direct proportion to the amount of stored fat. Both hormones can be transported through the BBB (Banks, 2008), and since they are not synthesized within the brain, intracerebral levels reflect serum levels. Once in the brain extracellular space, insulin and leptin signal through specific receptors expressed by many hypothalamic neurons (see below), and decrease feeding while increasing energy expenditure. Moreover, the sensory innervation from WAT projects to brain stem and hypothalamic circuits that regulate sympathetic outflow (Bartness, Shrestha, Vaughan, Schwartz, & Song, 2010). These neurons express leptin receptors and are thought to have a role in informing the brain on WAT lipid levels and lipolysis.

Both smoking (Reseland et al., 2005) and in vivo nicotine treatments can alter leptin levels, although these changes may be secondary to alterations of food intake and energy metabolism and, consequently, fat stores in adipocytes. In fact, chronic nicotine did not change leptin levels when food was restricted (Swislocki & Fakiri, 2008), increased leptin in comparison to pair-fed animals (Arai et al., 2001), and decreased leptin in free-feeding animals (Li & Kane, 2003). In addition, nicotine gum did not change plasma leptin levels in humans (Reseland et al., 2005). Confirming the notion that nicotine does not directly affect leptin release, nicotine exposure did not significantly affect leptin mRNA levels from cultured adipocytes and adipose tissue explants and leptin release from the explants (Reseland et al., 2005).

Smoking and nicotine exposure can induce nicotine resistance at the level of insulin target organs, and an effect may also be present at the level of insulin secretion. Smoking in humans as well as chronic nicotine treatment in rats decrease plasma insulin levels. Acute or prolonged nicotine exposure in pancreatic islets decreases basal or tolbutamide-elicited insulin release. This effect may be mediated by a direct action on α2, α3, α4, α5, α7, and β2 nAChRs expressed by β cells (Yoshikawa, Hellström-Lindahl, & Grill, 2005).

Overall, nicotine effects on humoral signals seem to be mediated primarily by other effects of nicotine on gastrointestinal function or food intake.

Neuronal Output

Contrary to the central nervous system (CNS), where nAChRs exert mostly modulatory actions, nicotinic transmission constitutes the backbone of neural information flow in two principal subdivisions of the peripheral motor nervous system, the motor somatic and the motor autonomic. We will not discuss in this context the peripheral motor somatic nervous system since muscle type nAChRs have low affinity for nicotine and are not influenced by the concentrations of nicotine used in animal experimentation in vivo and present in smokers.

nAChRs in Autonomic Motoneurons

Cholinergic neurons of the DMnX projecting to the gastrointestinal system express high levels of nAChRs at the somato-dendritic level (Sahibzada et al., 2002). In addition, nAChRs are expressed by presynaptic terminals of autonomic motoneurons where they positively regulate acetylcholine release (Coggan, Paysan, Conroy, & Berg, 1997). The composition of these receptors is in part similar to ganglionic neuron receptors, since moderate to high levels of α3, α4, α5, β2, β4, and α7 mRNAs are expressed in rodent DMnX (Allen Brain Atlas, http://www.brain-map.org). Pharmacological studies of the DMnX have shown that functional α7 receptors are expressed by DMnX motoneurons projecting to all gastrointestinal sectors and are coexpressed with α4β2 nAChRs in motoneurons projecting to the stomach and α3β4 nAChRs in motoneurons projecting to the intestine (Sahibzada et al., 2002).

The intermediolateral sympathetic motoneurons have been less intensively investigated. The intermediolateral column expresses binding sites for nicotinic agonists (Khan, Yaksh, & Taylor, 1994) with a pharmacological profile compatible with α4β2* and α3β4* subtypes, whereas α3/α6β2* and α7 nAChRs may not be expressed at high levels (Khan et al., 1994). Sympathetic motoneurons of this column express immunoreactivity for α3, α5, and β2 subunits (Khan et al., 2003). Presynaptic nAChRs on motoneuron terminals in sympathetic ganglia facilitate acetylcholine release. They are composed of β2* and α7 subtypes, whereas there is no functional evidence for β4* subtypes (Liang & Vizi, 1997).

nAChRs in Autonomic Ganglionic Neurons

nAChRs are the principal mediators of synaptic transmission in autonomic ganglia. Immunochemical and functional studies in rodents have shown that ganglionic neurons express high levels of α3β4 receptors in the somatic/dendritic compartment along with two minor populations of α3α5β4 and α3β2β4 receptors with different physiological and pharmacological properties (David et al., 2010; Mao et al., 2006). Low levels of α4* nAChRs may be expressed during development (Scholze et al., 2011). α7 receptors are also expressed, although they are perisynaptic (Berg & Conroy, 2002) and their contribution to ganglionic transmission in rodents may be modest (David et al., 2010). nAChRs are also expressed on presynaptic terminals of ganglionic neurons where they positively regulate noradrenaline or acetylcholine release (Fischer, Orr-Urtreger, Role, & Huck, 2005; Kristufek, Stocker, Boehm, & Huck, 1999). These receptors may be pharmacologically different from somatic/dendritic nAChRs (Fischer et al., 2005).

Overall, nAChRs stimulate sympathetic and vagal peripheral transmission in both motoneurons and ganglionic neurons. Nicotine may therefore mimic autonomic transmission in every respect, with multiple effects on the function of visceral organs that contribute to energy metabolism.

The Enteric Nervous System

The enteric nervous system is composed of about half a billion neurons (in humans, as many as in the spinal cord) organized into two intercommunicating webs: the myenteric and the submucousal plexus. This wide array of specialized (intrinsic primary afferent neurons, motoneurons, and interneurons) and richly heterogeneous neurons assures the functional autonomy of gastrointestinal system during digestion by regulating motility of gastrointestinal walls, digestive secretions, and local vascular tone (Furness, Jones, Nurgali, & Clerc, 2004). Most neurons in both the myenteric and submucosal plexus are cholinergic, with a high density of cholinergic synapses (and nAChRs, see below) in enteric circuits (Furness et al., 2004; Harrington, Hutson, & Southwell, 2010).

Similar to what occurs in the autonomic ganglia, enteric circuits predominantly involve nicotinic transmission, but the receptor subtypes and localization are much more complex. Overall, many enteric neurons are cholinergic (Furness et al., 2004; Harrington et al., 2010) and both electrophysiologically identified classes of enteric neurons, the S neurons (mostly motor neurons or interneurons) and the after hyperpolarizing (AH) neurons (mostly sensory neurons), express nAChRs because all excitatory postsynaptic potentials (EPSPs) recorded from enteric neurons are inhibited completely or partly by nicotinic antagonists (Galligan & North, 2004). Although the expression of nAChRs on function in these circuits is high, the prevalence of nicotinic transmission varies in different portions of the gastrointestinal tract and possibly in different species. For instance, it has been shown that all EPSPs are nicotinic in guinea pig stomach and rat colon, but only a portion of responses are nicotinic in guinea pig ileum (Galligan & North, 2004). Recent research has focused on the functional characteristics of presynaptic nAChRs on cholinergic motoneurons, which have an excitatory effect on acetylcholine release and therefore muscle contraction (Mandl & Kiss, 2007). These nAChRs may have major pharmacological importance, since they affect the endpoint of enteric integration directly.

The nAChR subtypes are less well characterized in the enteric system compared with similar studies in CNS and other ganglia. mRNA expression studies in postnatal rats, which are thought to reflect adult expression patterns, showed an autonomic ganglia-like expression pattern of subunits with relatively high levels of α3, α5, and β4 mRNA and lower levels of β2 mRNA in the myenteric plexus of stomach and intestine and relatively low levels of α3, α5, and β4 mRNA in the submucosal plexus of the intestine (Garza, Huang, Son, & Winzer-Serhan, 2009). A similar pattern was observed in immunohistological experiments with subunit-selective monoclonal antibodies (Obaid, Nelson, Lindstrom, & Salzberg, 2005). Accordingly a large predominance of β4* over β2* nicotinic binding was observed in receptor autoradiography studies (Garza et al., 2009). Electrophysiological evidence tallies well with expression studies, showing that 70% of AH neurons show typical α3β4-like responses, whereas the remaining AH neurons and S neurons have a different, and at present uncharacterized, nicotinic response (Galligan & North, 2004). Further electrophysiological studies with subtype-selective peptidic antagonists showed the presence of α3β2 and α3β2β4 responses (Obaid et al., 2005). Low levels of α7 mRNA are expressed in rat myenteric plexus (Garza et al., 2009) and both plexa of guinea pig (Obaid et al., 2005), although functional evidence for α7-like responses in enteric circuits is questioned (Galligan & North, 2004; Obaid et al., 2005).

From a functional standpoint, the role of the different nAChRs is difficult to assess in view of their presence at multiple levels of these complex circuits. In addition, nAChRs in both somatodendritic and presynaptic compartments stimulate both excitatory and inhibitory transmission, with potentially opposite functional outcomes. However, some basic insights have been obtained. In most species and experimental preparations, nicotine causes contraction of stomach, jejunal, and ileal muscles, and relaxation of colonic muscles and lower oesophageal sphincter (Braverman et al., 2011; Murakami et al., 2009; Vural, Ozturk Fincan, Bozkurt, Ercan, & Sarioglu, 2009). At a more integrated level, nAChRs are thought to play a crucial role in ascending reflexes but only a minor role in descending reflexes of the myenteric plexus, whereas in the submucosal plexus, most nAChRs are associated with descending projections (Galligan & North, 2004). Given this complexity, the effects of systemic nicotine treatment (and smoking) are variable with the dose of nicotine, a major determinant of the effect (Mandl & Kiss, 2007).

Adipose tissue

Brown Adipose Tissue

In rodents, developing humans, and possibly also adult humans, BAT is the principal mediator of adaptive thermogenesis (Enerback 2010; Wijers, Saris, & van Marken Lichtenbelt, 2009). BAT is activated by the sympathetic nervous system through β3 adrenergic receptors that turn on a cascade of intracellular events leading to the activation of uncoupling protein 1 (UCP1) activity. UCP1, in the inner mitochondrial membrane, causes proton leakage, and decreases the efficiency of the mitochondrial respiratory chain leading to heat production (Azzu & Brand, 2010). A recent paper demonstrated a second regulatory pathway for adaptive thermogenesis involving BAT-resident macrophages activated along the alternative pathway (see below) that can release noradrenaline locally (Nguyen et al., 2011). Mechanisms of adaptive thermogenesis in the skeletal muscle, the second tissue responsible for this function in adult humans, are not fully understood (Wijers et al., 2009).

Acute (Mano-Otagiri, Iwasaki-Sekino, Ohata, Arai, & Shibasaki, 2009; Yoshida, Sakane, Umekawa, & Kondo, 1994; but see Wellman, Marmon, Reich, & Ruddle, 1986 for opposite results) or chronic (Arai et al. 2001; Lupien & Bray, 1988; Yoshida, Yoshioka, Hiraoka, & Kondo, 1990; Yoshida et al., 1999) systemic nicotine as well as chronic cigarette exposure (Chen et al., 2008) increases sympathetic activity, noradrenaline release, and UCP1 expression in BAT, thus increasing thermogenesis. The target nAChRs for this effect are not known. Whereas there is no evidence for nAChRs in BAT (but see below the discussion of resident macrophages), both sympathetic ganglia and central sympathetic control circuits are possible candidates for nicotine effects on BAT. Systemic nicotine-elicited release of noradrenaline in BAT can be blocked by antagonists of the corticotropin-releasing hormone type 1 receptors (Mano-Otagiri et al., 2009), suggesting the involvement of hypothalamic control circuits of the sympathetic system (Usui et al., 2009).

White Adipose Tissue

As discussed above, a consistent effect of nicotine treatment is decreased body weight with reduction of WAT mass. Decrease in WAT mass may be due to a direct lipolytic effect. Both in vivo (Andersson & Arner, 2001; Mjos, 1988) and ex vivo (Sztalryd et al., 1996) nicotine, as well as smoking, (Laudon Meyer, Waldenlind, & Marcus, 2005) induces lipolysis in WAT. In addition, nicotine increases lipoprotein lipase activity in several tissues (Ashakumary & Vijayammal, 1997). These effects of nicotine lead to increased plasma levels of free fatty acids and other lipids (Ashakumary & Vijayammal, 1997). In parallel, nicotine increases free fatty acid utilization in muscle contributing to decreased adiposity (Sztalryd et al., 1996).

Nicotine’s lipolytic actions may depend on the repeatedly demonstrated effects on the sympathetic nervous system (see above), but also on a direct action on WAT. Experiments with local administration of nicotine through a microdialysis probe show that nicotine can stimulate lipolysis directly in WAT (Andersson & Arner, 2001) by an action on adipocytes or an indirect effect on other WAT cells. Nicotine increases lipolysis and inhibits fatty acid synthase (An et al., 2007) and lipoprotein lipase activity (Chajek-Shaul et al., 1994) in cultured adipocytes. Indeed, some evidence has been collected for intrinsic expression of nAChRs in WAT (Andersson & Arner, 2001). Cultured adipocytes express relatively low levels of nicotine binding (Liu, Mizuta, & Matsukura, 2004) with unknown composition, since a wide array of neuronal and muscle-type subunit mRNAs were detected with nonquantitative PCR (Liu et al., 2004). A second possible nicotine target could be WAT-resident macrophages.

Several lines of evidence point to the existence of a cholinergic anti-inflammatory pathway mediated by the activation of the vagus nerve and α7 nAChRs expressed by macrophages (Bencherif, Lippiello, Lucas, & Marrero, 2011). Activation of α7 leads to an anti-inflammatory orientation of macrophages, which involves suppression of genes of the classical pathway. Although not yet tested directly, it is likely that α7 activation leads to activation of macrophages along the alternative pathway, since α7-induced targets in macrophages, such as PI3 kinase (Rehani et al., 2008) and peroxisome proliferator–activated receptor-γ (Amoruso et al., 2007), are involved in cascades of the alternative pathway (de Keijzer et al., 2011; Martinez, 2011; Weisser et al., 2011). If true, this may be relevant for nicotine’s effects on BAT and WAT in view of the evidence for lipolytic and thermogenic effects of the macrophages activated along the alternative pathway (Nguyen et al., 2011). Moreover, this may be of pathophysiological relevance in obesity. Accordingly, nicotine acting on α7 receptors decreases WAT inflammation and improves glucose homeostasis and insulin sensitivity in genetic or diet-induced obese mice (Lakhan & Kirchgessner, 2011; Wang, Yang, Xue, & Shi, 2011).

Central Structures

A large number of studies have established that α4β2* and α7 nAChRs are the predominant heteromeric and homomeric subtypes, respectively, in the rodent CNS. For instance, histological studies in the main stations of the homeostatic system in the hypothalamus showed expression of α4 and β2 mRNAs in the paraventricular (PVH), ventromedial (VMH), and dorsomedial (DMH) hypothalamic nuclei and the lateral hypothalamus (LH) (Wada et al., 1989), whereas α7 mRNA is expressed in the arcuate nucleus (ARC), VMH, and DMH (Seguela, Wadiche, Dineley-Miller, Dani, & Patrick, 1993). In addition, there is growing evidence for the existence and functional relevance of uncommon subtypes in the CNS (see below). Moreover, most expression studies have been carried out in rodents. Primate CNS may have a wider variety of subtypes (Han et al., 2000).

Uncommon subtypes are present in two pathways of the reward circuits that express both high levels and a wide diversity of nAChR subtypes: the mesotelencephalic dopamine (DA) system and the habenulo-interpeduncular system.

Intensive research in recent years has established the main composition and functional features of nAChR subtypes in the mesotelencephalic DA neurons that are involved in drug, and perhaps food, reward: the four principal nAChR populations are α4β2, α4α5β2, α6β2β3, and α4α6β2β3 (De Biasi & Dani, 2011; Grady et al., 2007). These subtypes are expressed in variable proportions at the somatodendritic or terminal levels in different subregions of the brain, and in different species, with, for instance, α6* nAChRs more highly expressed in the rat than in the mouse, and even more highly in primates (Quik & McIntosh, 2006). In primates, expression of α3β2* and perhaps α2β2* nAChRs in the mesolimbic system has been proposed (Han et al., 2000; Quik & McIntosh, 2006). Finally, α7 receptors are expressed by about 40% of DA cell bodies but not by nerve terminals (Klink, de Kerchove d’Exaerde, Zoli, & Changeux, 2001), whereas α7 nAChRs are richly expressed by glutamatergic afferents projecting to the ventral tegmental area (VTA) and to the striatum (Jones & Wonnacott, 2004).

Although the lateral habenula expresses average levels of α4β2 nAChRs, the medial habenula and its target nucleus, the interpeduncular nucleus, in addition to α4β2 nAChRs, express α3β4, α3β3β4, α4β3β2, α2β2, and α4β4 nAChR subtypes (Grady et al., 2009). The medial habenular pathway is involved in nicotine’s aversive effects and in nicotine withdrawal syndrome (De Biasi & Salas, 2008) and may play a role in the hyperphagia observed during nicotine withdrawal.

In addition, uncommon subtypes may be expressed at low levels in a particular brain region but at a high level (and thus with an important functional role) in a specific cell type. A recently demonstrated instance is the expression of β4* nAChRs in pro-opiomelanocortin (POMC)-containing neurons (Mineur et al., 2011, see below). This underlines the necessity in CNS studies to uncover the precise localization and function of nAChR subtypes within the circuitry of interest.

Homeostatic System

The role of the homeostatic system is to maintain body weight (and more specifically, fat stores) over time, and in particular, to match energy (food) intake and energy expenditure (metabolic activity + physical exercise). To obtain this goal, the brain circuits of the homeostatic system regulate both meal features (size, quality, and length) and body energy metabolism.

The two main gates that allow signals related to meal, nutrient levels, and energy stores into the brain are the NTS (see above) and the ARC. The ARC also receives direct and indirect afferents from the NTS.

The ARC lies beside the ventral part of the third ventricle, adjacent to the median eminence, a brain region devoid of BBB through which hypothalamic releasing factors enter the blood circulation. Two neuronal subpopulations in the ARC (the GABAergic neuropeptide Y (NPY) and Agouti-related protein (AgRP)-containing neurons, and the POMC-containing neurons) constitute the main interface between peripheral signals and the brain (first-order neurons), representing the main central sensor of energy storage (Cone, 2005). Notably, AgRP and α-melanocyte-stimulating hormone (αMSH), the end product of POMC cleavage, share the same targets, melanocortin 3 and 4 receptors (MC3R, MC4R). Remarkably, AgRP is an antagonist and αMSH is an agonist of these receptors.

NPY/AgRP neurons, which express both leptin and ghrelin receptors, have low spontaneous activity and respond to fasting by increasing their activity. On the other hand, POMC neurons have high spontaneous activity. NPY/AgRP neurons project onto POMC neurons and release NPY (recognized by Y1 and Y2 receptors), AgRP, (recognized by MC3R) and GABA, thereby inhibiting POMC neuron activity. Leptin and ghrelin inhibit NPY neurons and activate POMC neurons. The inhibition of NPY neurons, in turn, reduces inhibition of POMC neurons, whereas the inhibition of POMC neurons results in a feedback loop leading to greater inhibition of these cells. Both NPY/AgRP and POMC neurons project to a wide number of hypothalamic and extra-hypothalamic brain nuclei on which they exert opposing effects, orexigenic (increase of food intake and decrease in energy expenditure) for NPY/AgRP neurons, and anorexigenic (decrease of food intake and increase in energy expenditure) for POMC neurons (Cone, 2005).

Within the hypothalamus, first-order neurons project to four main nuclei: the VMH, DMH, PVH, and LH, which are in turn strongly interconnected (Bouret, Draper, & Simerly, 2004; Sanchez-Lasheras et al., 2010; Williams et al., 2001). Their specific role can be summarized as follows: The VMH, the classical “satiety center,” projects mainly to brain stem locomotor areas and is a principal mediator of locomotor aspects of complex motivated behaviors, including feeding. In addition, recent evidence points to a principal role of VMH in food anticipation arousal (Ribeiro, LeSauter, Dupré, & Pfaff, 2009). The DMH is a crucial part of the food-entrainable oscillator that confers, together with the light-entrainable oscillator (controlled by the suprachiasmatic nucleus), circadian rhythmicity to feeding activity (Gooley, Schomer, & Saper, 2006; Mieda, Williams, Richardson, Tanaka, & Yanagisawa, 2006). The PVH mainly projects to autonomic control nuclei and to autonomic motoneurons and median eminence for releasing factor secretion. The PVH and DMH are thought to act as a single functional unit to initiate and maintain food intake, as well as to control energy expenditure. The LH, the classical “feeding center,” is composed of a web of neurons with very complex input and output connections. It contains, among others, two subpopulations of peptide-containing neurons (orexin and melanin-concentrating hormone, MCH), which are critically involved in feeding behavior, as both these peptides exert powerful orexigenic effects (Kalra et al., 1999). The LH sends strong projections to the brain stem and spinal cord, both to visceral sensory areas and autonomic and somatomotor output regions (Kerman, 2008; Kuwaki, 2011).

The hypothalamic homeostatic circuits are under the control of brain stem monoaminergic pathways. Increased DA transmission in the VMH is necessary to initiate meals (thereby regulating meal frequency), whereas its increase in the LH suppresses food intake (thereby regulating meal duration) (Meguid et al., 2000; Vucetic & Reyes, 2010). 5HT is a powerful negative modulator of food intake and its release during meals is thought to promote satiation. Noradrenaline transmission has different effects. Noradrenaline increases in PVH at the onset of the periods of food intake and stimulates carbohydrate intake through binding to α2 receptors (Alexander, Cheung, Dietz, & Leibowitz, 1993); however, noradrenaline in the NTS–PVH pathway is also a principal mediator of several satiation stimuli (Rinaman, 2011).

Nicotinic Receptor Expression and Function in Central Circuits of the Homeostatic System

First-order neurons

Given their core relevance for energy metabolism, first-order neurons have attracted a great deal of investigations as possible targets of nicotine. Chronic exposure of mice to low dose of cigarette smoke is paralleled by a decrease of NPY in PVH (Chen et al., 2007), despite reduced body weight and plasma leptin concentration. This suggests a direct inhibitory effect of smoke on NPY neurons that would therefore play a causal role in smoke-induced negative balance of energy metabolism. Initial studies with chronic continuous nicotine administration gave similar results, showing decreased NPY peptide levels, but not mRNA (Frankish et al., 1995) when compared with pair-fed rats. Other studies have resulted in opposite results, showing that chronic continuous treatment causes increase in NPY mRNA and peptide expression compared with pair fed rats (Li et al., 2000). Finally, subacute nicotine administration slightly increased NPY peptide but prevented the increase induced by food restriction (Jang et al., 2003). These results are difficult to reconcile and recall the limits of expression studies (see above). A recent paper on ex vivo preparations of ARC directly tested the electrophysiological effects of nicotine exposure on NPY neurons (Huang, Xu, & van den Pol, 2011). Nicotine at low doses directly causes NPY neuron depolarization, although it also reduced glutamatergic input to these neurons.

Nicotine exposure also has a direct excitatory effect on POMC neurons (Huang et al., 2011; Mineur et al., 2011), slightly stronger than that observed for NPY neurons (Huang et al., 2011). The functional role of these neurons in nicotine-elicited anorexigenic effect has been recently investigated with a combination of pharmacological, functional, and molecular genetic approaches. Systemic nicotine activates POMC neurons through an action on β4* nAChRs and reduces food intake. Genetic inactivation of POMC in the ARC or MC4Rs, the αMSH target receptors, in the PVH, prevents nicotine/cytisine-induced acute hypophagic effects (Mineur et al., 2011).

Lateral hypothalamus

LH neuronal circuits are possible targets of nicotine actions leading to a negative balance of energy metabolism. LH receives cholinergic afferents from basal forebrain, pontine nuclei, and local interneurons and expresses both α4β2 and α7 nAChRs (Seguela et al., 1993; Wada et al., 1989). Both systemic and intra-LH nicotine administrations alter the functioning of LH circuits.

It has been shown that acute systemic treatment with relatively high doses of nicotine induces Fos expression in selected cell subpopulations of the LH, comprising the orexin neurons but not the MCH neurons (Pasumarthi, Reznikov, & Fadel, 2006). This effect may be, at least in part, mediated by nicotine-elicited release of glutamate from prefronto-LH afferents and acetylcholine from basal forebrain–LH afferents (Pasumarthi & Fadel, 2010). On the other hand, ex vivo electrophysiological studies showed that nicotine can activate orexin neurons directly (Huang et al., 2011). Chronic treatment with high doses of nicotine increases orexin and orexin receptor mRNA and orexin peptide levels in the hypothalamus in comparison with pair-fed rats (Kane et al., 2000). Orexin binding was instead decreased with the same nicotine regimen (Kane, Parker, & Li, 2001). Moreover, orexin receptor 1 mRNA in LH was decreased at the end of nicotine self-administration (LeSage, Perry, Kotz, Shelley, & Corrigall, 2010). It seems unlikely that orexin neurons mediate the effects of acute or chronic nicotine exposure on energy metabolism, since orexin neurons are orexigenic. Indeed, it was shown that nicotine-sensitive orexin neurons project to the basal forebrain and paraventricular thalamic neurons and may therefore be implicated in effects of nicotine on arousal or cognition rather than food intake (Pasumarthi & Fadel, 2008).

In addition, nicotine withdrawal induced Fos expression in orexin neurons and in orexin target neurons in the PVH. Nicotine-elicited Fos induction in the PVH was blocked by systemic administration of orexin receptor antagonists. The functional consequence of orexin-mediated PVH activation upon nicotine withdrawal is the expression of withdrawal somatic signs, since these signs can be partially blocked by intra-PVH infusion of orexin antagonists (Plaza-Zabala, Martín-García, de Lecea, Maldonado, & Berrendero, 2010).

LH GABAergic transmission is thought to elicit decreases in food intake (Stanley, Urstadt, Charles, & Kee, 2011). Nicotine, as well as endogenous acetylcholine, facilitates GABA release in the LH through α7 receptors (Jo & Role, 2002; Jo, Wiedl, & Role, 2005). In turn, GABA release increases the inhibitory synaptic activity to the orexigenic MCH cells (Jo et al., 2005). Although the role of this mechanism in vivo has not been assessed directly, potentiation of LH GABA transmission followed by inhibition of orexigenic effectors may contribute to systemic nicotine-induced decreases in food intake.

Afferents to hypothalamic circuits

It has been hypothesized that nicotine-elicited release of monoamines in the hypothalamus (Fu, Matta, Brower, & Sharp, 2001; Meguid et al., 2000; Sharp & Matta, 1993; Sperlagh, Sershen, Lajtha, & Vizi, 1998; Yang, Blaha, Meguid, Oler, & Miyata, 1999) mediates its effects on food intake. Recent support for this hypothesis has been obtained with selective destruction of noradrenaline/adrenaline innervation of the LH. Animals devoid of adrenergic terminals become partially insensitive to nicotine-induced decreases in food intake and body weight (Kramer, Guan, Wellman, & Bellinger, 2007).

Another relevant target of nicotine-elicited catecholamine release in the hypothalamus is the anorexigenic CRH neurons. Both acute and chronic nicotine exposure activate CRH neurons of the PVH and the hypothalamic–pituitary–adrenal axis (Armario, 2010). This effect is not due to hypothalamic nAChRs but is thought to be mediated by nAChR-elicited release of glutamate from NTS visceral afferents, which in turn activates noradrenaline/adrenaline projections to the PVH (Zhao, Chen, & Sharp, 2007).

The Food Reward System

Palatable foods are rewarding and can be consumed even when animals are sated (i.e., in a manner that is partially independent from the requirements of energy homeostasis). It is commonly accepted that reward has multiple psychological/neurobiological components (Berridge & Robinson, 2003; Epstein, Leddy, Temple, & Faith, 2007). In this context, we will distinguish two main aspects of food reward: (a) reinforcement/motivation (“wanting”) and (b) hedonic content/affect (“liking”). Reinforcement represents the motivational, incentive, aspect of reward. Positive and negative reinforcers are stimuli that increase and decrease, respectively, the rate of a behavior that they follow. Hedonic content represents the affective, pleasurable, component of reward. Although these two aspects often go together, they can be separated and are thought to rely on partially independent neural substrates. Accordingly, palatable substances even devoid of caloric content are actively sought out, just like other rewarding substances and stimuli. Indeed, stimulation of food reward brain areas can trigger binge-like overeating even in sated animals. Current evidence indicates that palatable foods have strong motivational value that can override energy homeostatic systems. In fact, mechanisms and circuits mediating food reward are thought to be in part superimposable with those mediating drug addiction (Avena, Rada, & Hoebel, 2008; Kenny, 2011a). Palatable foods contain high levels of sugar and/or fat as well as calories. Overall, preference for palatable foods is driven by two food features: their sensory qualities (e.g., sweet taste) and their caloric content (so called post-ingestive effect) (de Araujo et al., 2008).

In rodents, sugar and fat are primary (unlearned or unconditioned) reinforcers (Epstein et al., 2007), and in humans, positive evidence has been obtained for sugar (Epstein, Carr, Lin, & Fletcher, 2011). Sensory cues associated with palatable foods (taste, texture, odour, and visual features), as well as behaviors associated with food consumption can become secondary (conditioned) reinforcers (Epstein et al., 2007). Moreover, taste processes, including those related to palatability, are influenced by nongustatory factors associated with food that include olfactory, somatosensory, and visceral signals.

Through the amygdala and orbitofrontal cortex (Rolls, 2006), sensory cues of palatable foods reach the corticolimbic and mesoaccumbens circuits implicated in brain reward. Core regions of these reward circuits in both rodents and humans comprise some cortical (limbic cortices, including orbitofrontal cortex and insula) and subcortical (amygdala, striatum, LH, dopaminergic ventral midbrain, and habenula) regions.

The functions of specific brain regions in the mediation of different aspects of food reward are starting to be understood. The mesotelencephalic, and particularly the mesolimbic, DA systems are perhaps the most intensely studied brain substrates for reward. Yet, the contribution of the mesolimbic system to food reward is not straightforward (Narayanan, Guarnieri, & DiLeone, 2010). Palatable foods or cues predicting palatable foods activate these neurons, and bitter tasting solutions inhibit them. However, mice with DA depletion in DA neurons show decreased feeding but maintain sucrose preference (Cannon & Palmiter, 2003). Accordingly, selective lesions of DA terminals in limbic target regions, including the nucleus accumbens (nAc), only transiently impair feeding (Narayanan et al., 2010). Overall, DA neurons appear to play a modulatory role in feeding and may be more involved in conditions when food procurement requires an effort or cues associated with food reward are presented, rather than in free feeding (Fulton, 2010; Kenny, 2011b; Narayanan et al., 2010). Moreover, mesolimbic DA neurons are a principal gate for the access of homeostatic signals to the reward circuits (see below).

Much recent research has demonstrated that a key node of food reward circuits is the nAc, the main target of the mesolimbic DA neurons as well as of pathways from the LH, the amygdala, and the limbic cortex. Striatal opioid systems are thought to regulate the hedonic properties of food. Infusion of mu opioid agonists into the nAc stimulates feeding, whereas infusion of mu opioid antagonists decreases consumption of preferred food (Kelley, Bless, & Swanson, 1996; Pecina & Berridge, 2005). The circuit underlying these effects has been described recently. Medium, spiny neurons of the food-related portions of nAc shell (that are inhibited by mu opioid receptors) exert a tonic inhibition on neurons of the anterior LH, in particular orexin neurons (that in turn project to midbrain DA neurons, see below, and lateral habenula) that drive feeding of palatable foods (Kenny, 2011b). The striato-hypothalamic pathway is considered a primary output of reward circuits to control food consumption.

Nicotine Effects on Central Circuits of the Reward System

Nicotine enhances several features of food reinforcement and hedonic content. Acute nicotine in rodents enhances positive taste reactivity, a test for palatability, of sucrose (i.e., food hedonic content). This effect disappears after chronic exposure to nicotine but may be reinstated after nicotine withdrawal (Parker & Doucet, 1995; see Wing et al., 2009 for partially different results). Nicotine also enhances the reinforcing properties of food and food-associated cues, that is, both primary and conditioned food reinforcers (Brunzell et al., 2006; Donny, Caggiula, Weaver, Levin, & Sved, 2011). Notably, this effect of nicotine is not associative (i.e., nicotine administration does not need to be contingent with the presentation of the reinforcer). For instance, mice exposed to acute nicotine had significantly higher breakpoints for sucrose during progressive ratio responding for food (Brunzell et al., 2006). When the rewarding power of food is decreased (for instance, in rich schedules of reinforcement such as free feeding), the nicotine effect is decreased. Although the subject has been less intensely investigated, available evidence suggests that chronic nicotine seems to cause the same enhancement of food reinforcement as acute nicotine (Donny et al., 2011). Interestingly, this effect remains after cessation of chronic treatment (Donny et al., 2011). Finally, nicotine increases the preference for smaller, sooner, compared with larger, later food rewards (a model for impulsive choice), and this effect is due to an insensitivity to reward magnitude (Locey & Dallery, 2011). Note that as for most nicotine effects, in general, there are rewarding effects of nicotine at doses around 0.1–0.5mg/kg and aversive effects at higher doses.

Although it is well known that nicotine powerfully affects brain reward systems (De Biasi & Dani, 2011), it is unknown what nAChR subtypes, cells, and circuits specifically mediate nicotine-enhancing effects on food reinforcers. An obvious candidate target is the DA mesotelencephalic system. Palatable foods or associated cues activate this system (see above) and nicotine is a powerful activator of DA neurons (De Biasi & Dani, 2011). Both acute and chronic systemic nicotine treatment elicit DA release from terminals of ventral and dorsal striatum. This effect is mediated by α4β2* and α6β2* nAChRs expressed at the somatodendritic or nerve terminal levels (De Biasi & Dani, 2011; Grady et al., 2007).

Nicotine’s effects are not limited to food reinforcers (Chaudhri et al., 2006; Donny et al., 2011). Voluntary exercise (and wheel running, its homologous in rodents) is considered a form of reinforcement (Garland et al., 2011). Nicotine markedly increases wheel running in rats habituated to the cage for several hours postinjection (Bryson, Biner, McNair, Bergondy, & Abrams, 1981; Gillman et al., 2008; but see Kuribara, Shinoda, & Uchihashi, 1995 for different effects in mice). When nicotine is administered chronically at fixed times of the day, increased wheel running is observed in the preinjection phase and persists for at least two days after cessation of the treatment (Gillman et al., 2008). When nicotine is administered multiple times a day, nicotine-elicited potentiation of wheel running is observed for a single injection/day and declines for the other times of administration (Gillman et al., 2010).

Nicotine is a psychomotor stimulant and has several other effects on locomotion. Although these effects are in part species specific, acute nicotine administration transiently (5–20min postinjection) increases locomotion in environments to which a rat has been habituated (Gotti et al., 2010), an effect that undergoes sensitization upon repeated administration (Mao & McGehee, 2010; Vezina, McGehee, & Green, 2007). Nicotine sensitization of habituated locomotion may be due to increased responsiveness to nicotine of the mesolimbic DA neurons (Vezina et al., 2007).

On the other hand, acute nicotine administration can decrease locomotion in a novel environment (Cohen, Welzl, & Bättig, 1991; Prus et al., 2008). Most important in the context of energy consumption are the data on spontaneous activity (SPA; Garland et al., 2011) over prolonged periods of time. Chronic continuous administration of nicotine increases SPA for the first two days of treatment followed by adaptation (Cronan, Conrad, & Bryson, 1985), though this effect has not been detected in all studies and may be dependent on sex and strain (Grunberg et al., 1985). Although early nicotine effects on locomotion are thought to depend on the activation of midbrain DA neurons (Gotti et al., 2010; Vezina et al., 2007), no data are available on the biological mechanisms mediating long-term effects on SPA.

Overall, the effects of nicotine on activity would increase caloric consumption (Garland et al., 2011) and would therefore likely decrease fat content and weight.

Interactions between Homeostatic and Reward Systems

The hedonic and reinforcing value of food is influenced by metabolic state (Kenny, 2011b). It is well known that nutritional status in rodents (as exemplified by conditions of food restriction or deprivation) modulates performance in behavioral paradigms in which motivation and reinforcement are involved, including behaviors in which food is the reward in rodents (Figlewicz, MacDonald Naleid, & Sipols, 2007) and human subjects (Kenny, 2011b).

The two principal adiposity signals, leptin and insulin, have powerful effects on food reward. Intracerebroventricular administration of low doses of insulin or leptin dampens the rewarding effect of palatable foods in rodent paradigms that test primary rewarding value, the association between food and the context where it is consumed and incentive value/motivation to work for a palatable food. These effects may be mediated, at least in part, by a direct action on reward circuits, namely the DA mesolimbic system (Kenny, 2011b; Narayanan et al., 2010). Receptors for leptin and insulin are expressed at high levels in midbrain DA neurons. Indeed, peripheral or local administration of leptin or insulin activates the respective intracellular cascades in VTA and decreases DA neuron firing as well as basal and food-elicited DA release in nAc (de Araujo, Ren, & Ferreira, 2010; Figlewicz et al., 2007; Fulton, 2010).

It has been shown recently that leptin (as well as fasting) modulates the taste-independent or postingestive (i.e., the effect due to caloric content) activation of DA neurons elicited by sucrose (Domingos et al., 2011). Principal meal control signals, such as CCK, PYY, and ghrelin (Fulton, 2010; Kenny, 2011b) modulate different aspects of the reinforcing effects of food on peripheral administration (Fulton, 2010). In particular, the effects of ghrelin may be due to a direct action on the VTA, where growth hormone secretagogue receptors are expressed. Accordingly, ghrelin infusions into the VTA or nAc increase DA neuron firing, DA release, and food intake (Abizaid et al., 2006; Andrews, 2011; Fulton, 2010).

In addition to direct actions of peripheral homeostatic signals on reward circuits, neuronal integration within homeostatic circuits influences food reward circuits. For instance, NPY and AgRP in the ARC not only increase food intake but also increase food reward for sucrose and fat, respectively (Fulton, 2010). The connection between LH, a principal component of the homeostatic circuits (see above), and the mesoaccumbens DA pathway plays an important role in the communication between homeostatic and reward circuits. LH neurons expressing two principal orexigenic peptides, orexin and MCH, project to the VTA and the nAc, respectively. Accordingly, orexin potentiates excitatory input to DA neurons and increases motivation for sucrose (Fulton, 2010; Thompson & Borgland, 2011), whereas MCH agonists and antagonists in the nAc increase and decrease, respectively, food intake (Fulton, 2010; Sears et al., 2010).

Relationship between Nicotine Effects on the Homeostatic and Food Reward System

Nicotine acts on central reward circuits to increase reinforcement and hedonic content of foods, thus leading to increased disposition to consume foods. As discussed above, in rodents freely fed on a standard lab chow, repeated or continuous nicotine administration decreases food intake and increases energy consumption. Since the rewarding value of reinforcers is decreased in conditions of easy access to the reinforcers themselves, such as free feeding, there appears to be nicotine-induced satiation and energy consumption that overcome the rewarding effects of nicotine under these conditions. This suggests that nicotine effects on the homeostatic system and the reward system have opposite outcomes.

In rodents freely fed on a high-fat diet, chronic nicotine, administered continuously or intermittently, maintains, and even increases, its effects on body weight and food intake compared with animals fed on a low-fat diet (Mangubat et al., 2012; Wellman et al., 2005), suggesting an increased prevalence of nicotine’s effects on the homeostatic system under these conditions. Contrasting data were obtained in mice with oral nicotine self-administration (Fornari et al., 2007). In these mice, an acceleration of weight gain paralleled by increased high-fat food intake was observed in the first weeks of nicotine treatment. Although this issue needs to be explored further, the modality of nicotine administration (e.g., active vs. passive) may alter the balance between nicotine effects on the homeostatic and reward systems.

It is well known that restricted access to a reinforcer enhances its reinforcing efficacy. Accordingly, nicotine enhancement of food reinforcement is more obvious in food-deprived animals and not under free-feeding conditions (see above, Donny et al., 2011). Unfortunately, the effects of chronic nicotine administration on body weight and food intake in food-restricted animals have not been carefully studied. The limited available evidence indicates that nicotine does not decrease weight in food-restricted rats (Swislocki, 2003, Swislocki & Fakiri, 2008). Accordingly, nicotine administration reduced caloric intake during or after a meal, but not in fasting conditions in human smokers (Perkins et al., 1991, 1992).

Another paradigm of interest in this context is the dissociation between palatability and caloric content of foods. Chronic continuous nicotine was administered to rats fed on four types of diet combining sweet/non-sweet and low/high-calorie foods. Nicotine-induced decrease in body weight was lowest in rats fed on sweet low calorie foods, suggesting that increased rewarding effect not coupled to increased caloric content increases the prevalence of reward with respect to homeostatic nicotine effects (Grunberg et al., 1985).

As discussed above, nicotine’s reinforcing effects persist during chronic treatment and for several days after treatment cessation. Donny and coworkers (2011) hypothesize that, on the contrary, nicotine enhancement of satiation is not maintained after treatment cessation. This would leave the enhancement of food reward by nicotine to predominate and could then cause the increase in food intake observed after nicotine withdrawal

Conclusions and Future Developments

Twenty to thirty percent of the population in most countries, including United States and European Union, smoke cigarettes and are exposed to nicotine. Therefore, the effects of nicotine on brain and body functions are of both medical and societal interest.

It has been shown repeatedly that smokers weigh less than nonsmokers (Albanes, Jones, Micozzi, & Mattson, 1987; Eisen, Lyons, Goldberg, & True, 1993) and that smoking cessation is accompanied by weight gain (Filozof et al., 2004; Klesges et al., 1997; Lycett, Munafo, Johnstone, Murphy, & Aveyard, 2011). In fact, smoking is used as a weight control strategy by many smokers, especially young women, and weight increase is perceived as a serious disincentive by many of those who wish to quit smoking (Clark et al., 2004; Klesges, Meyers, Klesges, & La Vasque, 1989; Klesges & Schumaker, 1992). Indeed, weight gain after smoking cessation can have health consequences, for instance, an increased incidence of diabetes (Yeh, Duncab, Schmidt, & Brancati, 2010).

Unfortunately, current pharmacologic smoking cessation therapies, such as nicotine replacement, varenicline and bupropion, have not been highly effective in controlling weight gain. As reviewed above, nicotine can decrease body weight in humans and in rodents with free access to food. Instead, varenicline does not have significant effects on food intake and body weight in human subjects (Stoops, Vansickel, Glaser, & Rush, 2008). In rodents, administration of varenicline reduced or increased responding for food, with pretreatment time a possible critical factor in determining this effect (Le Foll et al., 2011).

Preclinical studies have shown that bupropion causes mild weight loss in freely fed lean or obese rodents, possibly by increasing locomotor activity and thermogenesis (Billes & Cowley, 2007, 2008). Clinical trials have confirmed that bupropion causes long-term (1 year) weight loss when combined with lifestyle modification, an effect that appears independent from its antidepressant actions (Valentino, Lin, & Waldman, 2010).

When given as smoking cessation therapies, nicotine, varenicline, and bupropion all limited postcessation weight gain compared with placebo at the end of treatment though with moderate efficacy (Farley, Hajek, Lycett, & Aveyard, 2012). The efficacy of bupropion appears to be somewhat greater than that of nicotine and varenicline. However, the persistence of these effects in the longer term has not been verified. The plant alkaloid cytisine, a nicotinic agonist, has also been used as a smoking cessation aid in Eastern Europe; however, cytisine does not alleviate postcessation weight gain (Etter, 2006), unlike its effects in mice (Mineur et al., 2011), potentially because of differential effects on mouse and human nAChR subtypes.

The issue of nicotine actions on energy homeostasis has been the object of much research over the years. Yet, at the mechanistic level, there is much more that we do not know than what we know with some certainty. In this domain, even more than in others related to nicotine actions, the expression of nAChRs in most excitable cells and in many nonexcitable cells makes the problem of identifying the targets and mechanisms of nicotine effects on energy homeostasis objectively complex.

The literature reviewed here prompts a few final considerations. As a preliminary methodological consideration, the intrinsic complexity of neuronal central and enteric networks expressing nAChRs should be dealt with a more widespread application of innovative technologies. Molecular genetic approaches and the use of nAChR subtype-selective tools (antibodies and drugs) are at the moment the best available tools to dissect out the specific role of nAChRs in neuronal circuits.

With regard to what subjects would require future specific efforts, we would highlight a few issues. There is growing evidence that nAChRs are expressed and have a pathophysiological role in nonexcitable cells. The example of nAChRs in macrophages in WAT and BAT discussed above suggests that this issue needs a careful analysis also in the domain of energy homeostasis.

Another issue that has been relatively neglected despite its obvious relevance is a precise assessment of the role of nAChRs in the networks of the enteric nervous system, with respect to nAChR function not only in motility and secretion but also in information to the brain about appetite and satiation.

As amply reviewed above, nAChRs are highly expressed in neural networks detecting the taste and regulating the digestion of the different foods and nutrients, as well as in the neural circuits evaluating food palatability and its consequences on reward processes. Given the major difference in palatability and variety between experimental animal and human foods, a more intensive effort should be made to study in detail the role of nAChRs in consumption of animal fed diets containing different foods with varied nutrient composition and palatability.

The previous consideration introduces a core question that is currently largely unanswered, that is, what is the relationship between nicotine effects on the homeostatic and reward mechanisms regulating energy homeostasis? As discussed above, these effects are, broadly speaking, opposite and the former seems to prevail over the latter in free-feeding laboratory animals. Different dietary challenges as well as environmental and pathophysiological states should be explored both from a phenomenological and mechanistic standpoint to clarify this issue.

A final issue concerns the modality of nicotine administration. Most studies on nicotine effects are performed in acute or chronic passive administration paradigms. Neuroplastic changes elicited by nicotine self-administration have been shown to differ often from those elicited by passives administration. In view of its higher similarity with the human situation, animal models comprising active nicotine administration and eventual withdrawal should be explored in more detail.

Funding

M.Z. was supported by grants PRIN2009R7WCZS, Italian Ministry of Health RF‐2009‐1549619, and ERA-Net NICO-GENE. M.R.P. was supported by DA14241 from the National Institutes of Health.

Declaration of Interests

None declared.